Digital Synapse: From Architecture to Prediction
Our goal is to move from descriptive maps of the presynapse to experimentally testable predictions. We combine nanometer-scale molecular localization with quantitative measurements of neurotransmitter release, then integrate both layers in a computational model of the synapse.
Quantitative Mapping of Synaptic Protein Organization
me4Pi-SMLM uses multi-phase interferometric detection to localize presynaptic molecules with nanometer-scale precision. Compared with conventional three-dimensional STORM, the method sharpens axial localization and enables quantitative measurement of molecular distances within the active zone. This provides a route to map the relative organization of proteins such as Bassoon, GRM7 and CaV2.1, and to determine whether nanoscale architecture predicts release probability, vesicle priming and short-term plasticity. This work is developed in collaboration with Dr. Yongdeng Zhang and is presented in Yu, Zijing et al., Nature Biotechnology (2026).
A synapse viewed molecule by molecule
The rotating three-dimensional localization cloud reveals how two molecular populations occupy distinct but overlapping territories. These spatial relationships can be measured across many synapses and linked to functional phenotypes rather than treated as representative images alone.
From Molecular Architecture to Functional Prediction
A digital synapse integrates two complementary data layers. First, super-resolution imaging provides the absolute copy number and nanoscale position of active-zone, vesicle and regulatory proteins. Second, physiological assays quantify paired-pulse behavior, short-term plasticity, release probability, readily releasable pool size, replenishment and asynchronous release. A data-constrained model then connects molecular organization to synaptic output.
